[Wien] What is p-1/2 LOs ?

[Stefaan Cottenier]

>    You mean , for those which have 5p/6p as valence state, we dont need to
> use p-1/2.
> If I calculate PbTe with spin orbit, I only need to "put a zero in the 
> line ......" according to your last reply.Is it right?
>   But if i add p-1/2 for Pb,I can still get almost the same result. Does 
> it mean this result is not reasonable?

For Pb, a p-1/2 for the deep 5p is useful. For Te, a p-1/2 for the 4p is 
probably not contributing much. If you don't see much difference when adding 
p-1/2 for Pb, then it probably means it has not much effect in your case 
(which quantity do you compare?).

Don't forget that when using spin-orbit, you should test the convergence 
with respect to Emax. Adding p-1/2 can mean that you reach convergence at a 
much small Emax (3 instead of 10). See Kunes et al., PRB 64 (2001) 153102 
where this is demonstrated for the 6p in fcc Th.

Stefaan